ADVERTISEMENT

Bristol Myers Turns to Machine Learning Startup to Solve ALS

Bristol Myers Turns to Machine Learning Startup to Solve ALS

Bristol Myers Squibb Co. is teaming up with machine-learning startup Insitro in a novel $70 million deal to find new drugs for amyotrophic lateral sclerosis, the deadly illness also known as Lou Gehrig’s disease.

The five-year deal, which includes a $50 million upfront payment plus another $20 million in near-term incentives, is the latest sign that pharmaceutical companies think machine learning could untangle some of the most difficult-to-treat diseases.

If the research is successful in finding drug candidates, Insitro could receive more than $2 billion in discovery, development and commercial milestone payments as well as royalties on product sales.

“Machine learning and data generated by novel experimental platforms offer the opportunity to rethink how we discover and design novel medicines,” said Richard Hargreaves, a Bristol Myers senior vice president. “Insitro’s approach is novel and brings together machine learning and biology at scale.”

The deal is the latest in a series of drugmaker deals with artificial-intelligence startups. In September, Bayer AG inked a pact with Recursion Pharmaceuticals Inc. to use AI to find new drugs for lung fibrosis and other fibrotic diseases. The deal included a $30 million upfront payment and $50 million in equity funding by Bayer’s investment arm.

In the Insitro deal, the goal is to marry modern deep-learning methods with wet-lab science on a mass scale to glean insights that lead to new drug candidates for both ALS and frontotemporal dementia, another incurable brain disease and the most common form of dementia in people under age 60.

‘Poster Child’

Insitro already has a deal with Gilead Sciences Inc. to develop treatments for nonalcoholic steatohepatitis, a liver disease. It is backed with $243 million in venture funding from Arch Venture Partners, Andreessen Horowitz, mutual fund giant T. Rowe Price Associates Inc. and others.

“ALS is a poster child for what we hope to be able to achieve with machine learning,” said Daphne Koller, the founder and chief executive officer of Insitro. “The machine learning can really dig in and come up with significantly novel insights.”

Machine learning is well suited for complicated diseases such as ALS, where there are clues from genetics but the possible causes so numerous that it’s hard for the human mind to make sense of it, said Koller, a former Stanford University computer science professor and MacArthur “genius” grant recipient.

ALS causes motor neurons that control movement to slowly and inexorably die, leading to paralysis and death. While defective genes cause rare forms of the disease, researchers don’t know the cause of the vast majority of cases. There are few approved drugs, and pharmaceutical researchers aren’t sure how best to intervene.

That’s where the machine learning comes in. By using the computer to analyze massive amounts of data from motor neuron cells in the lab, some with various genetic defects and others derived from patients with sporadic cases, Koller’s team hopes to understand what types of experimental compounds can make the cells healthier, and under what circumstances.

If it works, “the machine learning will analyze those cells and figure out the profiles that correspond to healthy versus sick,” said Koller. “Then when you have that characterization, you can look for intervention that takes a cell that was sick and moves it over to a healthy state.”

©2020 Bloomberg L.P.